A Watershed-Scale Groundwater-Land-Surface ModelYuning Shi*, Kenneth Davis*, Christopher Duffy†, Xuan Yu†

*Department of Meteorology, the Pennsylvania State University† Department of Civil and Environmental Engineering, the Pennsylvania State University

Introduction

Model and data• A land-surface module is incorporated into the Penn StateIntegrated Hydrologic Model 2.0 (PIHM 2.0)–Fully coupled surface water, groundwater, and land-surface

components–Land-surface scheme is mainly adapted from the Noah LSM

• Susquehanna/Shale Hills Observatory (SSHO) in centralPennsylvania (0.08 km2)–Small-scale V-shaped catchment with 1st order stream–Real Time Hydrologic monitoring network (RTHnet) with an array

of land-surface and sub-surface sensors is installed in SSHO

Future Work• Questions to answer:–How does land-surface affect hydrologic predictions?–How does groundwater improve SEB prediction?

• Incorporate data assimilation module into model to assimilate in-situ measurements and optimize model parameters

• Study subsurface-land-surface interaction• Test model on different spatial scales• Evaluate model on flood/drought prediction at scales up to theJuniata River Basin (~8800 km2)

Land Surface Models- Ignore deep ground-water and lateral water flow

Groundwater models- Relatively simple evapo-transpiration scheme

Coupled models of groundwater and land surface - may yield significant improvements in short-term climate and flood/drought forecasting

Map of sensible heat flux (left) and latent heat fluxes(right) averaged over entire simulation period

Results

Simulated sensibleheat flux, latent heatflux, ground heat flux(G), and surface skintemperature (Tskin) asfunctions of watertable depth

Comparison of hourly water table depth (top left),discharge (bottom left), soil moisture content (bottomright) between model simulation and RTHnetmeasurements, and map of water table depth averagedover entire simulation period (top right)

• Simulation from 0000 UTC 01 May to 0000 UTC 01 Sep 2009• Model is spun-up by running from 01 May 2008 to 01 May 2009• Model is calibrated with in-situ measurements using “trial anderror” strategy

• Model time step is 1 minute and output interval is 1 hour

• Results could be improved by–applying better optimization method–using locally measured bedrock depth, soil map and

parameters, and vegetation cover and parameters–and adopting better physics

• Land surface variables are affected by topography, soil type, andlandcover type, and are correlated to groundwater table

Surface energy balance (SEB) predictions

Comparison of sensible heat flux (H), latent heat flux(LE), and net radiation (Rn) (from top to bottom) betweenmodel and RTHnet flux tower from 01 Aug to 01 Sep2009

• Goals:–Develop a fully-coupled groundwater-land-surface model–Comprehensively evaluate hydrologic and surface energy

balance predictions with high-frequency data at a measurement-rich site

Input data Source

Input data

Surface elevation Field surveySoil map and parameters SSURGOVegetation cover and parameters NLCD 2001Precipitation, air temperature, and RH RTHnet10-m wind speed, downward longwave radiation, downward solar radiation, and surface pressure

NARR

LAI, roughness length NLCD 2001

Evaluation data

Discharge RTHnet outlet gauge

Water table depth and soil moisture content RTHnet wells

Surface heat fluxes and net radiation RTHnet flux tower

Configuration of vegetation type(top left), soil type (top right) andsurface elevation (bottom) ofsimulation domain.

Grid setting for SSHO model domain.• Model domain–Total size 0.076 km2 with a triangular irregular network of 571

grids and 318 nodes–River channel represented by 21 river segments–Uniform bedrock depth (2 m)

• Model captures temporal variation of surface heat fluxesreasonably well

• Model performance is limited by quality of NARR radiationproducts

Hydrologic predictions

• Model captures temporal patterns, but tends to underestimateamplitudes in water table depth and soil moisture variation

• Model reproduces flood in June and low flow situations

Evaluation data

Input dataInput data

Model Predictions Calibration

Tuned parameters Model

Predictions

Evaluation dataEvaluationUntuned

parameters

Calibration

Evaluation